Compressor

Compressor

Dynamic Units

1 of 19

Compressor

Outline
Beginners Guide to Compression Theory.p3-p4

Compressor Settings. p5-p10

Compression Methods.p11-p14

Ableton Live 9 Compressor Unitp15-p16

Compressor Tips.p17-p18

Referencesp19

2 of 19

Compressor

Compressor Theory
The whole reason a compressor was originally introduced was to reduce the
dynamic range of a performance, which is particularly vital when working with
any form of music. Whenever you record any sound into a computer, sampler or
recording device, you should aim to capture the loudest signal possible so that
you can avoid artificially increasing the volume afterwards. This is because if you
record a source thats too low in volume and then attempt to artificially increase
it later, not only will it increase the volume of the recorded source, itll also
increase any background noise.
To prevent this, you need to record a signal as loud as possible but the problem
is that vocals and real instruments have a huge dynamic range. In other words,
the vocals, for example, can be quiet in one part and suddenly become loud in
the next (especially when moving from verse to chorus). Consequently, its
impossible to set a good average recording level with so much dynamic
movement since if you set the recording level to capture the quiet sections,
when it becomes louder the recording will go into the red clip. Conversely,
setting the recorder so that the loud sections do not clip, any quieter sections
will be exposed to more background noise.
Of course, you could sit by the recording fader and increase or decrease the
recording levels depending on the section being recorded but this would mean
that you need lightening reflexes. Instead, its much easier to employ a compressor to control the levels automatically. By routing the source sound through
a compressor and then into the recorder, you can set a threshold on the compressor so that any sounds that exceed this are automatically pulled down in
volume, thus allowing you to record at a more substantial volume overall.
A compressor can also be used to control the dynamics of a sound while mixing.
For example, a dance track that uses a real bass guitar will have a fairly wide
dynamic range, even if it was compressed during the recording stage. This will
cause problems within a mix because if the volume is adjusted so that the
loudest parts fit well within the mix, the quieter parts may disappear behind
other instrumentation. Conversely, if the fader is set so that quieter sections can
be heard over other instruments, the loud parts could be too prominent. Using
3 of 19

Compressor

compression more heavily on this sound during the mixing stage, the dynamic
range can be restricted, allowing the sound to sit better overall within the final
mix.
Although these are the key reasons why compressors were first introduced, it has
further, far-reaching applications for the dance musician and a compressors
action has been abused to produce the typical dance sound.
Since the signals that exceed the threshold are reduced in gain, the parts that
do not exceed the threshold arent touched, so they remain at the same volume as they were before compression. In other words, the difference in volume
between the loudest and the quietest parts of the recording are reduced, which
means that any uncompressed signals will become louder relative to the compressed parts. This effectively boosts the average signal level, which in turn not
only allows you to push the volume up further but also makes it sound louder

Waveform before compression

Waveform after compression

4 of 19

Compressor

Compressor Settings:

Threshold
The first control on a compressor is the threshold which when touched upon sets
the signal level where the compressor will begin squashing the incoming signal.
These are commonly calibrated in dB and will work in direct relation- ship with a
gain reduction meter to inform you of how much the compressor is affecting the
incoming signal. In a typical recording situation this control is set so that the
average signal level always lies just below the threshold, and if any exuberant
parts exceed it, the compressor will jump into action and the gain will be
reduced to prevent any clipping.
Ratio
The amount of gain reduction that takes place after a sound exceeds the threshold is set using a ratio control. Expressed in ratios, this control is used to set the
dynamic range the compressor affects, indicating the difference between the
signals entering the compressor that exceed the threshold to the levels that
come out of the other end.
For example, if the ratio is 4:1, every time the incoming signal exceeds the
threshold by 4dB, the compressor will squash the signal so that there is only a 1
dB increase at the output of the compressor. Similarly, if the ratio set is 6:1, an
increase at the compressors output of 1 dB will occur when the threshold is
exceeded by 6dB and likewise for ratios of 8:1, 10:1 and so on. Subsequently,
the gain reduction ratio always remains constant no matter how much
compression takes place. In most compressors these range from 1:1 up to 10:1
and may, in some cases, also offer infinity:1.
From this we can determine that if a sound exceeds a predefined threshold, the
compressor will squash the signal by the amount set with the ratio control. The
problem with this approach, however, is that we gain a significant amount of
information about sounds from their initial attack stage, and if the compressor
5 of 19

Compressor

jumps in instantaneously on an exceeded signal, it will squash the transients

which reduces its high frequency (HF) content.
For instance, if you are to set up a compressor to squash a snare drum, the
compressor will clamp down on the attack stage which in effect diminishes the
initial transients reducing it to a thunk. Whats more, this instantaneous action
will also appear when the sound drops below the threshold again as the compressor stops processing the audio. This can be especially evident when compressing low-frequency (LF) waveforms such as basses since compressors can
apply gain changes during a waveforms period.
In other words, if a low-frequency waveform, such as a sustained bass note, is
squashed the compressor may treat the positive and negative states of the
waveform as different signals and continually activate and deactivate. The result
of this is an unpleasant distortion of the waveform. To prevent this from
occurring, compressors will feature attack and release parameters.
Attack/Release
Both these parameters behave in a manner similar to those on a synthesizer but
control how quickly the volume is pulled down and how long it takes to rise back
to its nominal level after the signal has fallen below the threshold. In other
words, the attack parameter defines how long the compressor takes to reach
maximum gain reduction while the release parameter determines how long the
compressor will wait after the signal has dropped below the threshold before
processing stops.
This raises the obvious question that if the attack is set so that it doesnt clamp
down on the initial attack of the source sound, it could introduce distortion/
clipping before the compressor activates. While this is true, in practice very
short, sharp signals do not always overload an analogue recorder since these
usually have enough headroom to let small transients through without
introducing any unwanted artifacts. This isnt the case with digital recorders,
though, and any signals that are beyond the limit of digital can result in clipping,
so its quite usual to follow a compressor with a limiter or, if the compressor
features a knee mode, set it to use a soft knee.

6 of 19

Compressor

Soft/Hard Knee
All compressors will utilize either soft or hard knee compression but some will
offer the option to switch between the two modes. These are not controllable
parameters but dictate the shape of the envelopes curve, and hence the
characteristic of how the compressor behaves when a signal approaches the
threshold. So far weve considered that when a signal exceeds the threshold the
compressor will begin to squash the signal. This immediate action is referred to
as hard knee compression. Soft knee, on the other hand, continually measures
the incoming signal, and when it approaches 314dB (dependent on the
compressor) towards the current threshold, the compressor starts to apply the
gain reduction gradually.
Generally this will initially start with a ratio of 1:1, and as the signal grows ever
closer to the threshold, its gradually increased until the threshold is exceeded,
whereby full gain reduction is applied. This allows the compressors action to be
less evident and is particularly suitable for use on acoustic guitars and wind
instruments where you dont necessarily want the action to be evident.
It should be noted that the action of the knee is entirely dependent on the
compressor being used and some can be particularly long starting 12dB before
the threshold while others may start 3 dB before. As a matter of interest, 69dB
soft knees are considered to offer the most natural compression for instruments.
Peak/RMS
Not all compressors feature knees, so short transient peaks can sometimes catch
the compressor unaware and sneak past unaffected. This is obviously going to
cause problems when recording digitally, so many compressors will implement a
switch for Peak or RMS modes. Compressors that do not feature these two
modes will operate in RMS, which means that the compressor will detect and
control signals that stay at an average level rather than the short sharp transient
peaks. As a result, no matter how fast the attack may be set, theres a chance
that the transients will overshoot the threshold and not be controlled. This is
because by the time the compressor has figured out that the sound has
exceeded the threshold its too late the peaks been and gone again. Therefore
7 of 19

Compressor

to control short transient sound such as drum loops, its often prudent to engage
the peak mode. With this the compressor becomes sensitive to short sharp
peaks and clamps down on
them as soon as they come
close to the threshold,
rather than after they
exceed it. By doing so, the
peak can be controlled
before it overshoots the
threshold and creates a
problem.
While this can be particularly useful when working with drum and percussion
sounds it can create havoc with most other timbres. Keep in mind that many
instruments can exhibit a particularly short, sharp initial attack stage, and if the
compressor perceives these as possible problems, itll jump down on them
before they overshoot. In doing so, the high-frequency elements of the attack
will be dulled which can make the instrument appear less defined, muddled or
lost within the mix. Therefore, for all instruments bar drums and percussion, its
advisable to stick with the RMS mode.
Make-Up Gain
The final control on a compressor is the make-up gain. If youve set the threshold, ratio, attack and release correctly, the compressor should compress
effectively and reduce the dynamics in a sound but this compression will also
reduce the overall gain by the amount set by the ratio control. Therefore,
whenever compression takes place you can use the make-up gain to bring the
signal back up to its pre-compressed volume level.
Side Chaining
Alongside the physical input and output connections on hardware compressors,
many also feature an additional pair of inputs known as side chains. By inputting
an audio signal into these, a sounds envelope can be used to control the action
that the compressor has on the signal entering the normal inputs. A good
example of this is when a radio DJ begins talking over a record and the volume
8 of 19

Compressor

of the record lowers so that their voice becomes audible, then when they stop
speaking the record returns to its original volume. This is accomplished by
feeding the music through the compressor as normal but with the microphone
connected into the side chain. This supersedes the compressors normal
operation and uses the signal from the side chain rather than the
threshold as the trigger. Thus, the compressor is triggered when the microphone is spoken into, compressing (in effect lowering the volume of the music)
by the amount set with the ratio control. This technique should only be viewed
as an example to explain the process, though, and more commonly side
chaining is usually used to make space in a mix for the vocals.
In a typical mix, the lead sound will occupy the same frequencies as the human
voice, resulting in a cluttered mid-range if the two are to play together. This can
be avoided if the lead mix is fed into the main inputs of the compressor while
the vocal track is routed into the side chain. With the ratio set at an appropriate
level (dependent on the tonal characteristics of the lead and vocals) the lead
track will dip when the vocals are present, allowing them to pull through the mix.
STANDARD COMPRESSION
Despite the amount of control offered by the average compressor, they are
relatively simple to set up for recording audio. As a generalized starting point,
its advisable to set the ratio at 4:1 and lower the threshold so that the gain
reduction meter reads between 8 and 10 dB on the loudest parts of the signal.
After this, the attack parameter should be set to the fastest speed possible and
the release set to approximately 500ms. Using these as preliminary settings, they
can then adjusted further to suit any particular sound.
Its advisable that compression is applied sparingly during the recording stage
because once applied it cannot be removed. Any exuberant parts of the
performance should be prevented from forcing the recorders meters into the
red while also ensuring that the compressor is as transparent as possible. Solidstate compressors are more transparent than their valve counterparts and so are
better suited for this purpose.

9 of 19

Compressor

As a general rule of thumb, the higher the dynamic range of the instrument
being recorded the higher the ratio and the lower the threshold settings need to
be. These settings help to keep the varying dynamics under tighter control
and prevent too much fluctuation throughout the performance. Additionally, if
the choice between hard or soft knee is available, the structure of the timbre
should be taken into account. To retain a sharp, bright attack stage, hard knee
compression with an attack setting that allows the initial transient to sneak
through unmolested should be used, provided of course that the transient is
unlikely to bypass the compression. In these instances, and to capture a more
natural sound, soft knee compression should be used.
Finally, the release period should be set as short as possible but not so short
that the effect is noticeable when the compressor stops processing. After setting
the release at 500 ms, the time should be continually reduced until the processing is noticeable and then increased slowly until it isnt.
Some compressors feature an automode for the release that uses a fast release
on transient hits and a slower time for smaller peaks, making this task easier.
As a more practical example, with a simple four to the floor kick running through
the compressor and the ratio and threshold controls set so that the gain
reduction reads 8dB on each kick, its necessary to ensure that the gain
reduction meter returns to 0 dB during any silent periods. If it doesnt, then the
loop is being overcompressed. If the gain reduction only drops to 2 dB during
the silence between kicks, then it makes sense that only 6 dB of gain reduction
is actually being applied. This means that every time the compressor activates it
has to jump from 0 to 8dB, when in reality it only needs to jump in by 6dB. This
additional 2dB of gain will distort the transient that follows the silence, making it
necessary for the gain reduction to be adjusted accordingly.

10 of 19

Compressor

Compression Methods

Variable MU
The first compressors ever to appear on the market were called variable MU
units. This type of compressor uses valves for the gain control circuitry and does
not have an adjustable ratio control. Instead of an adjustable control, the ratio is
increased in proportion to the amount of the incoming signal that exceeds the
threshold. In other words, the more the level overshoots the threshold the more
the ratio increases. While these compressors do offer attack and release stages,
theyre not particularly suited towards material with fast transients, even with
their fastest attack settings. Due to the valve design, the valves run out of
dynamic range relatively quickly, so its unusual to acquire more than 1520 dB
of gain reduction before the compressor runs out of energy. Nevertheless,
variable MU compressors are renowned for their distinctive, phat, lush character,
and can work magic on basses and pumping dance mixes. The most notorious
variable MU compressors are made by Manley and can cost in excess of 3500.
FET
FET compressors use a field effect transistor to vary the gain. These were the
first transistors to emulate the action of valves. They provide incredibly fast
attack and release stages making them an excellent choice for beefing up kick
and snare drums, electric guitars, vocals and synthesizer leads. While they suffer
from a limited dynamic range, if theyre pushed hard they can pump very
musically and are perfectly suited for gain pumping a mix. The only major
problem is getting your hands on one. Original FETs are as rare as rocking horse
manure, and consequently second-hand models are incredibly expensive.
Reproduction versions of the early FETs, such as the UREI 1176LN Peak Limiter

11 of 19

Compressor

(approximately 1800) and the LA Audio Classic II (approximately 2000), are a

worthwhile alternative.
Optical
Optical (or opto) compressors use a light bulb that reacts to the incoming
audio by glowing brighter or dimmer depending on the incoming sound
(seriously!). A phototransistor tracks the level of illumination from the bulb and
changes the gain. Because the phototransistor must monitor the light bulb
before it takes any action, some latency is created in the compressors response,
so the more heavily the compression is applied the longer the envelope times
tend to be. Consequently, most optical compressors utilize soft knee
compression. This creates a more natural attack and release but also means that
the compressor is not quick enough to catch many transients. Despite this,
optical compressors are great for compressing vocals, basses, electric guitars
and drum loops, providing that a limiter follows the compression. (Limiters will
be explained later.)
There are plenty of opto compressors to choose from, including the ADL 1500
(approximately 2500), the UREI LA3 and UREI Teletronix LA-2A (approximately
2900 each), the Joe Meek C2 (approximately 250) and the Joe Meek SC2.2
(approximately 500). Both Joe Meek units sound particularly smooth and warm
considering their relatively low prices, and for the typical gain pump
synonymous with dance then you could do worse than to pick up the SC2.2.
Notably, all Joe Meeks compressors are green because after designing his first
unit he decided to spruce it up by coloring it with car aerosol paint and green
was the only color he could find in the garage at the time.
Computer-Based Digital
Computer-based digital compressors are possibly the most precise compressors
to use on a sound. Because these compressors are based in the software
domain, they can analyze the incoming audio before it actually reaches the
compressor, allowing them to predict and apply compression without the risk of
any transients sneaking past the compressor. This means that they do not need
to utilize a peak/RMS operation. These digital compressors can emulate both
solid-state, transparent compression and the more obvious, warm, valve
12 of 19

Compressor

compression at the fraction of the price of a hardware unit. In fact, the Waves
RComp can be switched to emulate an optical compressor. Similarly, the PSP
Vintage Warmer and Sonalksis TBK3 can add an incredible amount of valve
warmth.
The look-ahead functions employed in computer-based compressors can be
emulated in hardware with some creative thought, which can be especially
useful if the compressor has no peak function. Using a kick drum as an example,
make a copy of the kick drum track and then delay it in relation to the original by
50ms. By then feeding the delayed drum track into the compressors main inputs
and the original drum track into the compressors side chain, the original drum
track activates the compressor just before the delayed version goes through the
main inputs, in effect creating a look-ahead compressor!
Ultimately, it is advisable not to get too carried away when compressing audio as
it can be easy to destroy the sound while still believing that it sounds better.
This is because louder sounds are invariably perceived as sounding better than
those that are quieter. If the make-up gain on the compressor is set at a higher
level than the inputted signal, even if the compressor was set up by your pet cat,
it will still sound better than the non-compressed version. The incoming signal
must be set at exactly the same level as the output of the compressor so that
when bypassing the compressor to check the results, the difference in volume
doesnt persuade you that it sounds better.
Furthermore, while any sounds that are above the threshold will be reduced in
gain, those below it will be increased when the make-up gain is turned up. While
this has the advantage of boosting the average signal level, a compressor does
not differentiate between music and unwanted noise. So 15dB of gain reduction
will reduce the peak level to 15 dB while the sounds below this remain the same.
Using the make-up gain to bring this back up to its nominal level (i.e. 15dB) any
signals that were below the threshold will also be increased by 15dB, and if
there is noise present in the recording, it may become more noticeable.
Most important of all, dance music relies heavily on the energy of the overall
punch produced by the kick drum, which comes from the kick drum physically
moving the loudspeakers cone in and out. The more the cone is physically
13 of 19

Compressor

moved, the greater the punch of the kick. This degree of movement is directly
related to the size of the kicks peak in relation to the rest of the musics
waveform. If the difference between the peak of the kick and the main body of
the music is reduced too much through heavy compression, it may increase the
average signal level but the kick will not have as much energy since the dynamic
range is restricted, meaning that all the music will move the cone by the same
amount. So, you should be cautious as to how much you compress otherwise
you may lose the excursion which results in a loud yet flat and unexciting track
with no energetic punch from the kick.

14 of 19

Compressor

Ableton Live 9 Compressor Unit

Compressor can be switched between three basic modes of operation. One not
covered in this handbook is the Expand Mode. In Expand mode, the Ratio can
be set to values below 1. In this state, Compressor acts as an upward expander,
and will increase the gain when signals exceed the threshold.
The opposite of compression is expansion. A typical expander lowers the levels
of signals that are below a threshold. This is how Lives Gate works, and is more
accurately called downward expansion because it pushes quiet signals down,
thus increasing the dynamic range. It is also possible to increase a signals
dynamic range by raising the levels of signals that are above a threshold. Like
upward compression, this technique is known as upward expansion and is much
less common. This diagram shows that either type of expansion results in a
signal with a larger dynamic range.

15 of 19

Compressor

To summarize:

Downward compression (common): make loud signals quieter

Upward compression (uncommon): make quiet signals louder
Downward expansion (common): make quiet signals quieter
Upward expansion (uncommon): make loud signals louder
The Multiband Dynamics device allows for all of these types of processing. In
fact, because the device allows for incoming audio to be divided into three
frequency bands, and each band has both an upper and lower threshold, a
single instance of Multiband Dynamics can provide six types of dynamics
processing simultaneously.

Envelope Follower
In addition to these modes, Compressor can be switched between two envelope
follower shapes that offer further options for how the device measures and
responds to signal levels. In linear (Lin) mode, the speed of the compression
response is determined entirely by the Attack and Release values. In logarithmic
(Log) mode, sharply compressed peaks will have a faster release time than less
compressed material. This can result in smoother and less noticeable
compression than Lin mode. Note that the Lin/Log switch is not visible in
Compressors collapsed view.
*Both the modes are unique to Ableton Live 9 and are not featured in Lives predecessor
version. Instead you are presented with three envelope follower shapes FF1/FF2 (forwardfeeding) and FB (feedback).

16 of 19

Compressor

Compressor Tips
This section presents some tips for using Compressor effectively, particularly
with the sidechain options.
Mixing a Voiceover
Sidechaining is commonly used for so-called ducking effects. For example,
imagine that
you have one track containing a voiceover and another track containing
background music. Since you want the voiceover to always be the loudest
source in the mix, the background music must get out of the way every time the
narrator is speaking. To do this automatically, insert a Compressor on the music
track, but select the narration tracks output as the external sidechain source.
Sidechaining in Dance Music
Sidechaining/ducking is a dance music producers secret weapon because it can
help to ensure that basslines (or even whole mixes) always make room for the
kick drum. By inserting a com- pressor on the bass (or master) track and using
the kick drums track as the sidechain input, you can help to control problematic
low frequencies that might interfere with the kick drums attack.
Using the sidechain EQ in conjunction with this technique can create ducking
effects even if you only have a mixed drum track to work with (as opposed to an
isolated kick drum). In this case, insert the Compressor on the track you want to
duck. Then choose the drum track as the external sidechain source. Then enable
the sidechain EQ and select the lowpass filter. By carefully adjust- ing the
Frequency and Q settings, you should be able to isolate the kick drum from the
rest of the drum mix. Using the sidechain listen mode can help you tune the EQ
until you find settings youre happy with.
Note that mastery of these techniques may result in a noticeable increase in
drink tickets, remix offers and dates.

17 of 19

Compressor

Common Compressor Settings

The settings shown in the table above are naturally only starting points and too
much compression should be avoided during the recording stage, something
that can only be accomplished by setting both the ratio and the threshold
control carefully. This involves setting the compressor to squash audio but
ensuring that it stops processing and that the gain reduction meter drops to 0
dB (i.e. no signal is being compressed) during any silent passages.

18 of 19

Compressor

References:
p1-p12_Rick Snoman_The Dance Music Manual_Second Edition
p13-15_Ableton Live 9 Official Manual

19 of 19